A Terragrunt frontend that allows execution of Terragrunt/Terraform through Docker.
Table of content:
TGF
is a small utility used to launch a Docker image and automatically map the current folder, your HOME folder and your current environment
variables to the underlying container.
By default, TGF is used as a frontend for our fork of terragrunt, but it could also be used to run different endpoints.
Using TGF
ensures all your users are using the same set of tools to run infrastructure configuration even if they are working on different environments (linux
, Microsoft Windows
, Mac OSX
, etc).
Terraform
is very sensitive to the version used and if one user updates to a newer version, the state files will be marked with the latest version and
all other users will have to update their Terraform
version to the latest used one.
Also, tools such as AWS CLI
are updated on a regular basis and people don't tend to update their version regularly, resulting in many versions
among your users. If someone makes a script calling a new feature of the AWS
api, that script may break when executed by another user that has an
outdated version.
You can install TGF
using the provided GitHub Action:
steps:
- uses: coveooss/tgf@master
On mac
and linux
:
You can run the get-latest-tgf.sh
script to check if you have the latest version of tgf installed and install it as needed:
curl https://raw.githubusercontent.com/coveooss/tgf/master/get-latest-tgf.sh | bash
On Windows
, run get-latest-tgf.ps1
with Powershell (version 7.x or more):
(Invoke-WebRequest https://raw.githubusercontent.com/coveooss/tgf/master/get-latest-tgf.ps1).Content | Invoke-Expression
This will install tgf in your current directory. Make sure to add the executable to your PATH.
TGF has multiple levels of configuration. It first looks through the AWS parameter store
under /default/tgf
using your current AWS CLI configuration if any. There, it tries to find parameters called config-location
(example: bucket.s3.amazonaws.com/foo) and config-paths
(example: my-file.json:my-second-file.json, default: TGFConfig). If it finds config-location
, it fetches its config from that path using the go-getter library. Otherwise, it looks directly in SSM for configuration keys (ex: /default/tgf/logging-level
).
Note: The SSM configuration will only be read if AWS environment variables are set, the AWS CLI is installed or the ~/.aws folder exists. If you wish to force TGF to read the SSM config and these conditions are not met, you can set the TGF_USE_AWS_CONFIG=true
environment variable
TGF then looks for a file named .tgf.config or tgf.user.config in the current working folder (and recursively in any parent folders) to get its parameters. These configuration files overwrite the remote configurations. Your configuration file could be expressed in YAML or JSON.
Example of a YAML configuration file:
docker-refresh: 1h
logging-level: notice
Example of a JSON configuration file:
{
"docker-refresh": "1h",
"logging-level": "notice"
}
Key | Description | Default value |
---|---|---|
docker-image | Identify the docker image to use | coveo/tgf |
docker-image-version | Identify the image version | |
docker-image-tag | Identify the image tag (could specify specialized version such as k8s, full) | latest |
docker-image-build | List of Dockerfile instructions to customize the specified docker image) | |
docker-image-build-folder | Folder where the docker build command should be executed | |
docker-refresh | Delay before checking if a newer version of the docker image is available | 1h (1 hour) |
docker-options | Additional options to supply to the Docker command | |
logging-level | Terragrunt logging level (only applies to Terragrunt entry point). Critical (0), Error (1), Warning (2), Notice (3), Info (4), Debug (5), Full (6) |
Notice |
entry-point | The program that will be automatically launched when the docker container starts | terragrunt |
tgf-recommended-version | The minimal tgf version recommended in your context (should not be placed in .tgf.config file ) |
no default |
recommended-image | The tgf image recommended in your context (should not be placed in .tgf.config file ) |
no default |
environment | Allows temporary addition of environment variables | no default |
run-before | Script that is executed before the actual command | no default |
run-after | Script that is executed after the actual command | no default |
alias | Allows to set short aliases for long commandsmy_command: "--ri --with-docker-mount --image=my-image --image-version=my-tag -E my-script.py" |
no default |
auto-update | Toggles the auto update check. Will only perform the update after the delay | true |
auto-update-delay | Delay before running auto-update again | 2h (2 hours) |
update-version | The version to update to when running auto update | Latest fetched from Github's API |
Note: The key names are not case-sensitive
It is possible to specify configuration elements that only apply on a specific os.
Example of an HCL configuration file:
docker-refresh: 1h
logging-level: notice
windows:
logging-level: debug
linux:
docker-refresh: 2h
section | Description |
---|---|
windows | Configuration that is applied only on Windows systems |
linux | Configuration that is applied only on Linux systems |
darwin | Configuration that is applied only on OSX systems |
ix | Configuration that is applied only on Linux or OSX systems |
> tgf -H
usage: tgf [<flags>]
DESCRIPTION:
TGF (terragrunt frontend) is a Docker frontend for terragrunt/terraform. It automatically maps your current folder,
your HOME folder, your TEMP folder as well of most environment variables to the docker process. You can add -D to
your command to get the exact docker command that is generated.
It then looks in your current folder and all its parents to find a file named '.tgf.config' to retrieve the
default configuration. If not all configurable values are satisfied and you have an AWS configuration, it will
then try to retrieve the missing elements from the AWS Parameter Store under the key '/default/tgf'.
Configurable values are:
- docker-image
- docker-image-version
- docker-image-tag
- docker-image-build
- docker-image-build-folder
- docker-image-build-tag
- logging-level
- entry-point
- docker-refresh
- docker-options
- recommended-image-version
- required-image-version
- tgf-recommended-version
- environment
- run-before
- run-after
- alias
- update-version
- auto-update-delay
- auto-update
Full documentation can be found at https://github.com/coveooss/tgf/blob/master/README.md
Check for new version at https://github.com/coveooss/tgf/releases/latest.
Any docker image could be used, but TGF specialized images could be found at: https://hub.docker.com/r/coveo/tgf/tags.
Terragrunt documentation could be found at https://github.com/coveo/terragrunt/blob/master/README.md (Coveo fork).
Terraform documentation could be found at https://www.terraform.io/docs/index.html.
ENVIRONMENT VARIABLES:
Most of the arguments can be set through environment variables using the format TGF_ARG_NAME.
Ex:
TGF_LOCAL_IMAGE=1 ==> --local-image
TGF_IMAGE_VERSION=2.0 ==> --image-version=2.0
SHORTCUTS:
You can also use shortcuts instead of using the long argument names (first letter of each word).
Ex:
--li ==> --local-image
--iv=2.0 ==> --image-version=2.0
IMPORTANT:
Most of the tgf command line arguments are in uppercase to avoid potential conflict with the underlying command.
If any of the tgf arguments conflicts with an argument of the desired entry point, you must place that argument
after -- to ensure that they are not interpreted by tgf and are passed to the entry point. Any non conflicting
argument will be passed to the entry point wherever it is located on the invocation arguments.
tgf ls -- -D # Avoid -D to be interpreted by tgf as --debug
It is also possible to specify additional arguments through environment variable TGF_ARGS.
VERSION: 1.23.1
AUTHOR: Coveo
Flags:
-H, --help-tgf Show context-sensitive help (also try --help-man)
--image=coveo/tgf Use the specified image instead of the default one
--image-version=version Use a different version of docker image instead of the default one
-T, --tag=latest Use a different tag of docker image instead of the default one
--local-image If set, TGF will not pull the image when refreshing
--get-image-name Just return the resulting image name
--refresh-image Force a refresh of the docker image
-E, --entrypoint=terragrunt Override the entry point for docker
--current-version Get current version information
--all-versions Get versions of TGF & all others underlying utilities
-L, --logging-level=<level> Set the logging level (panic=0, fatal=1, error=2, warning=3, info=4, debug=5, trace=6, full=7)
-D, --debug Print debug messages and docker commands issued
-F, --flush-cache Invoke terragrunt with --terragrunt-update-source to flush the cache
--interactive ON by default: Launch Docker in interactive mode, use --no-interactive to disable
--docker-build ON by default: Enable docker build instructions configured in the config files, use
--no-docker-build to disable
--home ON by default: Enable mapping of the home directory, use --no-home to disable
--temp ON by default: Map the temp folder to a local folder (Deprecated: Use --temp-location host and
--temp-location none), use --no-temp to disable
--temp-location=folder Determine where the temporary work folder 'tgf' inside the docker image is mounted:
volume: Mounts the work folder in the docker volume named “tgf”. The volume is created if it doesn't exist.
host: Mounts the work folder in a directory on the host.
none: The work folder is not mounted and is private to the docker container.
--mount-point=<folder> Specify a mount point for the current folder
--prune Remove all previous versions of the targeted image
--docker-arg=<opt> ... Supply extra argument to Docker
--with-current-user Runs the docker command with the current user, using the --user arg
--with-docker-mount Mounts the docker socket to the image so the host's docker api is usable
--ignore-user-config Ignore all tgf.user.config files
--aws ON by default: Use AWS Parameter store to get configuration, use --no-aws to disable
-P, --profile=<AWS profile> Set the AWS profile configuration to use
--ssm-path=<path> Parameter Store path used to find AWS common configuration shared by a team
--config-files=<files> Set the files to look for (default: TGFConfig)
--config-location=<path> Set the configuration location
--update Run auto update script
Example:
> tgf --current-version
tgf v1.22.0
Returns the current version of the tgf tool
> tgf -- --version
terragrunt version v1.2.0
Returns the version of the default entry point (i.e. Terragrunt
), the --version located after the -- instructs tgf to pass this argument
to the desired entry point
> tgf -E terraform -- --version
Terraform v0.11.8
Returns the version of Terraform
since we specified the entry point to be terraform.
- Terraform
- Terragrunt
- Go Template
- Shells & tools
sh
openssl
All tools included in coveo/tgf:base
plus:
- Python (2 and 3)
- Ruby
- AWS CLI
- jq
- Terraforming
- Tflint
- Terraform-docs
- Terraform Quantum Provider
- Shells
bash
zsh
fish
- Tools & editors
vim
nano
zip
git
mercurial
All tools included in coveo/tgf
plus:
All tools included in coveo/tgf:aws
plus:
kubectl
helm
All tools included in the other images plus:
- AWS Tools for Powershell
- Oh My ZSH
- Shells
powershell
> tgf plan
Invoke terragrunt plan
(which will invoke terraform plan
) after doing the terragrunt
relative configurations.
> tgf apply -var env=dev
Invoke terragrunt apply
(which will invoke terraform apply
) after doing the terragrunt
relative configurations. You can pass any arguments
that are supported by terraform
.
> tgf plan-all
Invoke terragrunt plan-all
(which will invoke terraform plan
on the current folder and all sub folders). Terragrunt
allows xxx-all operations to be
executed according to dependencies that are defined by the dependencies statements.
> tgf -E aws s3 ls
Invoke AWS CLI
as entry point and list all s3 buckets
> tgf -E fish
Start a shell fish
in the current folder
> tgf -E my_command -i my_image:latest
Invokes my_command
in your own docker image. As you can see, you can do whatever you need to with tgf
. It is not restricted to only the pre-packaged
Docker images, you can use it to run any program in any Docker images. Your imagination is your limit.